This specification relates to annotating electronic resources with geographic information.
Electronic resources (e.g., web pages, documents, pictures, videos, and so on) can be associated with certain locations. The strength of the association can be quantified by assigning a score to an association between a resource and a location. For example, a web page for the New York Subway is likely more related to New York than it would be to Los Angeles. Additionally, that web page is typically more interesting to residents of and visitors to New York than it would be to people in Los Angeles, although it could be useful for someone in Los Angeles planning a trip to New York. Thus, the strength of association between the web page for the New York Subway and New York is stronger than with the location of Los Angeles.
The degree to which a resource is associated with, or related to, a particular location can be conventionally determined and quantified by direct analysis of the resource. For example, for the web page for the New York Subway, the number of appearances of the word “New York” can be counted and used to determine a relatedness score for the page. Similarly, techniques that mine keywords from images and videos can be used to attempt to determine the relatedness of those resources to locations.